Micro-electromechanical system (MEMS) mirrors (or micro-mirrors) have been evolving for approximately two decades as part of the drive toward integration of optical and electronic systems, for a range of uses including miniature scanners, optical switches, and video display systems. These structures consist of movable mirrors fabricated by micro-electronic processing techniques on wafer substrates (for example silicon, glass, or gallium arsenide). The torsional micro-mirror typically comprises a mirror and spring assembly suspended over a cavity formed in or on a base. The mirrors are electrically conductive, as is at least one region behind the mirror, affixed to the stationary base, so that an electric field can be formed between the mirror and the base. This field is used to move the mirror with respect to the base. An alternative comprises the use of magnetic materials and magnetic fields to move the mirrors.
Typically the mirror surface consists of either the wafer itself or a deposited layer (metal, semiconductor, or insulator), and generally in the prior art the springs and mirror are formed from the same material (but not in all cases). The mirror and torsion springs are separated from the base by an etch process, resulting in the formation of a cavity between the mirror and base.
For display or image acquisition applications, the goal is to develop compact systems with rapid frame rates (at least 60 Hz) and high resolution, consisting of between 200 and 2000 miniature pixels per line. For scanning system designs in this range, the mirrors should be large (in the range of 200 .mu.m x 200 .mu.m to 2 mm x 2 mm), fast (in the range of between 3 kHz and 60 kHz for resonant devices), and scan a photon beam through a large angle 7 to 40 degrees).
In optical systems that contain very small elements, diffraction by the smallest element may introduce diffraction broadening and deleteriously increase the final pixel size.
Enlarging the limiting element reduces this broadening and militates for larger mirrors. However, as mechanical systems get larger (for example, increasing the size of a torsional mirror), they tend to be characterized by greater mass and consequently lower resonant frequency; this resonant frequency sets the scanning speed of the system. A frequency in the range of 5 to 50 kHz is desirable. Prior art mirror designs have been limited by the difficulty inherent in obtaining a high resonant frequency with a large mirror size, free from diffraction broadening effects. In prior art cases in which the mirror mass is made very low to obtain high resonant frequency, the resultant reduction in stiffness of the mirror is a limiting factor in the quality of the reflected image. This problem is exacerbated by the possibility of heating of the mirror by light absorbed in the mirror. Such heating militates for a thick mirror capable of conducting the heat away from the source.
The scanning angle through which the mirror moves determines the number of distinguishable pixels in a display or imaging system. Therefore, a large scanning angle is desirable. Generally in the prior art the scan angle is limited by the presence of electrodes that interfere with mirror motion (but not in all cases).
Electrostatic actuation is the most common method used to drive micro-mirrors. In order to produce a force, a voltage is generated between two electrodes, usually the plates of a parallel plate capacitor, one of which is stationary and the other of which is attached to the mirror as described previously. By making the mirror an electrical conductor, the mirror itself can be made to serve as one of the plates. The force generated for a given voltage depends on the plate area and on the gap between the plates, which may change as the mirror position changes. For torsional mirrors, the important drive parameter is the torque, and the effective torque on the structure is also proportional to the distance between the resultant force and the axis of rotation of the mirror. Thus, a large driving force can be achieved using large capacitor plates and small gaps; by applying the force at a distance from the rotation axis, a large torque may be obtained.
In many prior art designs the criteria for a large deflection angle range tend to be in conflict with the criteria for large driving forces. The deflection angle is limited by the presence of surfaces behind the mirror. An example of a limiting surface would be the bottom of a cavity in the base etched beneath the mirror, or some other substrate on which the mirror is mounted. The maximum angle is achieved when the mirror contacts this backplane, so the small separation between the mirror and the backplane needed for generating adequate electrostatic deflection force limits the maximum angle. Accordingly, in prior art designs in which the mirror is used as one of the drive electrodes and the other electrode is on the backplane, increasing the gap reduces the force or torque obtained at a given voltage. Some prior art designs use electrodes that are offset from the main mirror body and which are connected through actuator linkages, allowing the backplane to be moved further away or even eliminated entirely. Typically, though, these electrodes have smaller active areas and shorter moment arms, which tend to reduce the effective forces and torques as well. Additionally, if as the mirror moves, the gap between the drive electrodes narrows, then the gap still may be a limiting factor for the range of motion of the structure.
A second set of design problems arises in the selection of the mirror. Prior art designs and processes do not permit the mirror to be made from very low mass material without also sacrificing structural rigidity. One of the process limitations is the use of the same material for torsion spring and mirror mass, or the same set of patterning steps for spring and mirror mass. The selection of mirror materials with a view toward the elastic or fatigue properties of the springs restricts the suitability of the material with respect to mirror mass rigidity, and also limits the optical performance of the mirrors.
In 1980, Peterson disclosed a silicon torsional micromachined mirror (U.S. Pat. No. 4,317,611; K. E. Peterson, "Silicon torsional scanning mirror," IBM J. Res. Dev., 24(5), 1980, pp. 631-637). Both the mirror and torsion elements were patterned in a thin (134 microns) silicon wafer and retained the full thickness of the wafer. The structure was then bonded to a glass substrate, over a shallow well to allow room for the mirror motion. Actuation of the device was electrostatic. The mirror body was used as one electrode and the other electrodes were placed at the bottom of the well under the mirror. A narrow ridge in the well under the axis of rotation of the mirror was used to eliminate transverse motion of the structure. The manufacturing process for this device was relatively simple, requiring a single patterning step for the silicon and two patterning steps for the glass substrate. Its resonance frequency was about 15 kHz, and at resonance the angular displacement reached about 1.degree.. The limitations of this device are related to the depth of the well. A 2 mm mirror touches the bottom of a 12.5 .mu.m well at a displacement of 0.7.degree. (1.4.degree. total motion). Increasing the well depth to increase the range of motion is not necessarily desirable, because it proportionally reduces the torque achieved for a given voltage.
Nelson (U.S. Pat. No. 5,233,456), Baker et al (U.S. Pat. No. 5,567,334), Hornbeck (U.S. Pat. No. 5,552,924), and Tregilgas (U.S. Pat. Nos. 5,583,688 and 5,600,383) have developed and patented a series of torsional mirror designs and improvements for use in deformable mirror device (DMD) displays. These mirrors are fabricated by surface micromachining, consisting of a series of patterned layers supported by an undisturbed substrate. The DMD display uses an individual mirror at each pixel. The mirrors are therefore designed to be very small, to be operated in a bi-stable mode, and to maximize the packing fraction on the surface of the display. To minimize the gaps between the reflecting surfaces of adjacent mirrors, the support structure and drive components are fabricated in underlying layers, requiring a complicated multi-step deposition and patterning process. As with the Peterson mirror, the Hornbeck mirror is designed to serve as one of the deflection electrodes, and the others are placed behind the mirror. Owing to the small size of the mirrors (about 20 .mu.m x 20 .mu.m), high deflection angles are attainable with reasonably small gaps. These mirrors are designed for driving at low frequencies, and for significant dwell at a given angle (on or off), rather than for continuous motion, although the early development included mirrors designed for resonant operation (U.S. Pat. No. 5,233,456). A scanned display or imager requires, however, a large mirror, and the difficulties with scaling up torsional mirrors that are driven electrostatically with plates mounted behind the mirrors prevent the Hornbeck mirrors from being easily modified for use in scanning display applications.
Toshiyoshi describes a silicon torsion mirror for use as a fiber optic switch (H. Toshiyoshi and H. Fujita, "Electrostatic micro torsion mirrors for an optical switch matrix," J. Microelectromechanical Systems, 5(4), 1996, pp. 231-237). The Toshiyoshi mirror is a relatively large device (400 .mu.m on a side and 30 .mu.m thick), which rotates about an axis close to one edge of the mirror. The mirror is defined by etching the silicon wafer from the front, and the excess wafer material is etched from the back of the wafer. It is thus suspended over a cavity in the wafer, supported by very thin (0.3 .mu.m) metal torsion rods. The structure is then bonded onto another substrate, on which electrodes have been plated. Toshiyoshi has demonstrated separation of the mechanical properties of the springs and mirror by using silicon for the mirror mass, and metal for the springs. Actuation is electrostatic, by placing a voltage between the mirror body and the electrodes of the lower substrate. The range of motion is limited by the mirror hitting the glass substrate, at about 30.degree.. In order to obtain the maximum deflection at an applied voltage of 80 volts, the stiffness of the torsion members must be very low, achieved by making them very thin. This also limits the resonant frequency of the structure to 75 Hz, making the approach unsuitable for a scanned display or scanned imager. Thus Toshiyoshi has not shown how the separation of the mechanical properties of the spring and mirror can be used to attain a high resonant frequency and high angular displacement.
Dhuler of the MCNC has disclosed a mirror wherein the mirror body is formed from the silicon substrate, while the supports and actuators are fabricated above the mirror plane using surface micromachined polycrystalline silicon layers (V.J. Dhuler, "A novel two axis actuator for high speed large angular rotation," Conference Record of "Transducers '97," 1997). The mirror body is first defined using ion implantation of boron as an etch stop, and then by removal of the excess Si wafer from the back of the mirror. The supports and drive electrodes are offset from the top surface of the substrate by posts, which define the gap between the drive capacitor plates. Thus the mirror is free to rotate unhindered by the bottom surface of a well, while the drive torque, being applied by actuators, is not limited by a requirement for a large capacitor gap. While it represents a significant advance in the state of the art, this device suffers from certain flaws which the current invention resolves.
In the MCNC process the mirror body thickness is limited by the boron implantation process, which has limited penetration depth; the disclosed mirror was 4 .mu.m thick. The stiffness of the mirror is limited by both its size and thickness, so larger mirrors need to be thicker to avoid deformation of the mirror surface in use. For scanning applications, flexure in the mirror leads to uncertainty in the pixel size and location and distortion of the pixel shape. The implantation process also introduces stress into the mirror body, causing deformation of the reflective surface. The supports and actuators of the MCNC device are formed in a multi-step process and, as they are non-conducting, require the separate deposition and patterning of electrodes.
Kiang describes a 200 .mu.m x 250 .mu.m mirror that has a frequency of 15 kHz and maximum displacement of 15.degree. (M. H. Kiang, "Surface micromachined electrostatic comb driven scanning micromirrors for barcode applications," 9th Annual Workshop on Micro Electro-Mechanical Systems, 1996, San Diego, Calif., pp. 192-197). This mirror is made of deposited and patterned surface layers, and before using it must be first rotated out of the plane of the substrate using a comb drive and locked into position using complicated hinges. This approach obviates the problem of forming a cavity behind the mirror. However, the use of surface micromachined layers means that the structural rigidity of the micro-mirror cannot be controlled (because the thickness is limited to thin (a few microns) layers). The mirror motion is obtained by electrostatic drive applied by an actuator linked to one edge of the mirror. The motion of the mirror is restricted by the actuation mechanism.
Other torsional micromirrors are mentioned in the literature (M. Fischer, "Electrostatically deflectable polysilicon torsional mirrors." Sensors and Actuators, 44(1), 1996, pp. 372-274; E. Mattsson, "Surface micromachined scanning mirrors," 22d European Solid State Device Research Conference, Sep. 14-17, 1992, vol. 19, pp. 199-204). Most are small (less than 100 .mu.m on a side) and have very small displacements, not suitable for scanning applications. The exceptions tend to be complicated to fabricate or actuate and suffer from the same shortcomings as the mirrors described above.
Magnetically actuated cantilevered MEMS mirrors have been disclosed by Miller et al. of the California Institute of Technology (R. Miller, G. Burr, Y.C. Tai and D. Psaltis, "A Magnetically Actuated MEMS Scanning Mirror," Proceedings of the SPIE, Miniaturized Systems with Micro-Optics and Micromachining, vol. 2687, pp. 47-52, January 1996; R. Miller and Y.C. Tai, "Micromachined electromagnetic scanning mirrors," Optical Engineering, vol. 36, no. 5, May 1997). Judy and Muller of the University of California at Berkeley disclosed magnetically actuated cantilevered structures which may be used to support mirrors (Jack W. Judy and Richard S. Muller, "Magnetic microactuation of polysilicon flexure structures," Journal of Microelectromechanical Systems, 4 (4), December 1995, pp. 162-169). In both cases, the moving structures are supported by cantilever beams along one edge. They are coated with a magnetic material, and upon the application of a magnetic field at an angle to the mirror surface, the mirror rotates in the direction of the field, bending the cantilevers. Miller has also disclosed a similar mirror which uses a small coil fabricated on the moving structure to provide it with magnetic moment. In Miller's mirror, the springs are formed out of the original silicon wafer, and in Judy's mirror the springs are fabricated out of a polysilicon layer deposited for the purpose. The conduction path for the magnetic coil device is provided by a separate NiFe contact.